结合机械设计约束处理的差分进化算法

针对机械设计中的约束优化问题,提出了改进约束处理的自适应罚函数法。结合一般机械约束优化问题维数不高和差分进化算法简单、高效的特点,应用差分进化算法容易求得机械约束优化问题的全局最优解。给出了2个机械约束优化的数值实例,与已有的文献结果比较,表明新方法处理机械约束优化问题稳健且有效。     

工程约束优化的自适应罚函数混合离散差分进化算法

将离散约束优化问题转化为非负整数约束规划问题,开发求解该问题的离散差分进化算法。该算法采用基于混沌映射的种群初始化、双版本变异和带随机扰动项的取整运算等新策略。针对非线性约束条件,给出惩罚基数的计算方法和连续映射基函数的表达式,在此基础上设计处理非线性约束的自适应惩罚因子。提出一种刻画种群多样性的新测度——种群二次平均基因距离及基于新测度的依概率混沌移民算子。将自适应罚函数法、依概率混沌移民操作与离散差分进化算法有机融合,构造面向工程约束优化的混合离散差分进化算法。对3个离散约束优化实例进行验证,结果表明,混合算法具有良好的鲁棒性且优于离散粒子群算法。应用混合算法求解斜齿圆柱齿轮传动优化设计问题,结果优于遗传算法及其改进算法、离散粒子群算法,目标函数值较遗传算法及其改进算法分别下降41%和10%。     

基于自适应变异差分进化算法的电弧时域模型

随着电弧炉功率不断增大,电弧炉对供电网络的负面影响越来越受到关注.主要的负面影响是引起谐波、电网电压波动和闪变.因此,建立精确的电弧模型,对于研究上述问题有着重要的意义.新的电弧时域模型以能量守恒定律为基础,用非线性微分方程描述电弧电导与电流之间的函数关系,并且利用现场检测到的数据,采用自适应变异差分进化算法对参数进行辨识.通过调整参数,模型完全可以模拟电弧炉冶炼过程的电弧特性.仿真结果表明,模型输出的电压、电流与现场实测数据一致,验证了模型的正确性.     

基于离散差分进化算法的齿轮优化设计

根据机械设计中大多数优化变量属于混合变量的实际情况,提出采用基于离散变量方法解决机械优化设计问题。开发了一种不同变量自由设置间距的离散差分进化算法,连续变量采用合适的微小间距,离散变量采用工程规定间距。结合机械约束处理的自适应罚函数方法,以典型机械产品———齿轮和齿轮轴重量最轻优化实例为例,进行数值实验。结果表明该算法可以非常方便地处理约束优化问题,与粒子群算法实验对比,具有较高精度和可靠性。离散变量优化结果无需作圆整后处理,避免了圆整处理带来的系列问题。     

自适应差分进化改进云资源调度

为了解决云制造资源调度与实际制造场景不符、易陷入局部最优和收敛速度过慢的问题,提出了一种基于自适应多目标差分进化的制造资源调度方法。该方法考虑了实际云制造平台资源调度特点,建立了一个具有时序约束及成本约束的多目标优化资源调度模型。针对云制造资源调度特点,对传统差分进化进行改进,提出自适应变异率和交叉率,实现了寻优过程中变异率与交叉率的动态调整,均衡了多目标差分进化全局搜索与局部搜索能力,提高了最优解搜索的精度与速度。实际算例证明了该方法的有效性和可行性。     

逃逸离散差分进化算法在齿轮传动优化中的应用

根据决策变量映射关系,将齿轮传动设计中的离散约束优化问题转化为约束非负整数规划问题(Constrained non-negative integer programming problems,CNIPPs),并应用离散差分进化(Discrete differential evolution,DDE)算法求解该问题。引入定量评价种群多样性的平均基因距离指标,并据此提出一种采用反向学习算子生成新个体的自适应逃逸策略,以克服基本DDE算法求解离散问题易陷入局部最优区域的缺点。将逃逸策略融入DDE算法,并结合可行性规则约束处理技术,形成求解CNIPPs的逃逸离散差分进化(Escape DDE,EDDE)算法。应用EDDE算法求解齿轮传动优化设计实例,并提出用于比较多种算法优化性能的相对综合性能指标。通过测试与分析可知,新算法具有良好稳健性和可靠性,且综合指标优于对比算法。优化结果明显好于已有文献的最优解,齿轮质量下降了27%。     

并联机构位置正解的自适应差分进化算法

根据杆长约束条件,建立求6自由度一般6-SPS并联机构位置正解的无约束优化模型,再应用差分进化(Differential evolution,DE)算法求解该问题。针对基本DE算法可能出现进化停滞或陷入局部极值区域的缺点,提出一种引入新个体的自适应策略,以增强算法全局优化性能。将引入新个体的自适应策略融入DE算法,并使用混合变异算子及基于三角函数扰动的缩放因子和交叉因子,形成自适应差分进化(Adaptive DE,ADE)算法。数值结果表明,对于一般6-SPS并联机构正运动学分析问题,ADE算法能以较少计算开销求出全部高精度位置正解。通过与基本DE算法、自适应变异粒子群算法和改进人工蜂群算法比较,验证了ADE算法的收敛精度和计算稳健性指标优于对比算法。     

基于新自适应差分进化算法的Magic Formula轮胎模型参数辨识方法

针对传统差分进化法固定的控制参数设置与进化策略难以适应复杂多变的问题的状况,提出一种新的自适应差分进化方法,并将其应用于魔术公式(Magic formula,MF)轮胎模型的参数辨识中以解决其参数辨识难的问题。该方法结合基于成功进化个体的控制参数选择策略以及基于双审判矢量的进化策略,实现控制参数的有效自适应。通过对纯侧偏工况下轮胎数据的侧向力和回正力矩参数辨识,证明该方法比另外两种先进的自适应差分进化算法IOA和SspDE具备更好的全局优化与快速收敛能力,也比传统的数值优化Levenberg-Marquardt方法识别精度更高,是辨识MF轮胎模型参数的一种有效手段。     

融合自适应邻域差分进化的FastSLAM算法

机器人导航与建图中FastSLAM算法应用最广,但FastSLAM算法存在较多的问题,对算法中存在粒子集退化、重采样中丢失粒子多样性问题进行研究,研究结果表明,融合了自适应邻域差分进化算法的AMNDE-FastSLAM算法能有效改善这一问题。AMNDE-FastSLAM算法根据FastSLAM算法中的运动模型对粒子进行位姿估计更新,从建议分布中提取粒子,将FastSLAM算法中的重采样步骤替换为具有自适应能力的改进邻域差分进化算法。最后的实验结果表明,与传统的FastSLAM算法相比,AMNDE-FastSLAM算法具有更好的鲁棒性,在处理机器人被"绑架"问题上收敛速度更快、更优。     

基于遗传和差分进化算法的备件库存协同控制模型

分析了基于供应成本优化的备件协同控制库存决策模型,为获得全局收敛能力强、速度快,尤其是稳定可靠的新算法,设计了基于遗传和差分进化算法的混合智能求解算法,并通过两个实例与遗传算法、标准的差分进化算法进行了性能对比,显示出了所提算法的优势.基于所设计的算法,分别对用户需求、运输成本和最晚供应到达时间等参数进行了敏感性分析,讨论了数据不确定性对库存协同控制的影响程度,进而给出了不同环境下的供应方式优化建议.     

模糊资源约束的联合补充问题

针对不确定环境下的多产品联合补充问题,用三角模糊数表示不确定的资源约束,建立了模糊规划模型,目标函数为最小化订货成本和库存持有成本,决策变量为基本补充周期和每种产品的补充周期.用遗传算法对模型进行求解,以模糊规划模型的目标函数值作为染色体的适应度,阐述了染色体编码、选择、交叉、变异等遗传操作.最后,给出了仿真数值实例,比较了模糊资源约束模型和确定资源约束模型对1 600个随机生成问题的计算结果.     

Model and algorithm for fuzzy joint replenishment and delivery scheduling without explicit membership function

We study a joint replenishment and delivery scheduling (JRD) problem in which a central warehouse serves n-retailers in the presence of vague operational conditions such as ordering cost and inventory holding cost. In the proposed fuzzy set-based approach, an exact membership function is not assumed and instead can be approximated using piecewise linear functions based on alpha level sets because of their easy handling and efficiency. Subsequently, the fuzzy total cost is defuzzified by the widely used signed distance method to ranking fuzzy numbers. However, due to the JRD's difficult mathematical properties, efficient and effective solution procedures for the problem have eluded researchers. To find an optimal solution, an effective and efficient differential evolution (DE) algorithm is designed. After determining the appropriate parameters of the DE by parameter tuning test, the effectiveness of the DE is verified by numerical examples. We compare the DE with the available best approach and results show that DE can solve this non-deterministic polynomial hard problem in a robust way with a high convergence rate and low average error.     

A genetic algorithm to optimize multiproduct multiconstraint inventory control systems with stochastic replenishment intervals and discount

There are two main assumptions in multiperiodic inventory control problems. The first is the continuous review, where, depending on the inventory level, orders can happen at any time, and the other is the periodic review, where orders can only happen at the beginning of each period. In this paper, these assumptions are relaxed, and the periods between two replenishments are assumed independent and identically distributed random variables. Furthermore, the decision variables are assumed integer-type and that there are two kinds of space and budget constraints. The incremental discounts to purchase products are considered, and a combination of backorder and lost sales are taken into account for the shortages. The model of this problem is shown to be a mixed integer-nonlinear programming type, and in order to solve it, both genetic algorithm and simulated annealing approaches are employed. At the end, two numerical examples are given to demonstrate the applicability of the proposed methodologies in which genetic algorithm method performs better than simulated annealing in terms of objective function values.     

Projection data replenishment algorithm for limited angle tomography

A new iterative algorithm of tomographic reconstruction of objects on the basis of projection data available in a limited range of angles only is proposed. The algorithm is based on calculating artificial projections in those directions where projection data are unavailable. By means of numerical simulations, it is verified that the algorithm developed ensured high quality of reconstruction up to the angular interval of 45–60°.     

Minimizing transportation cost of a joint inventory location model using modified adaptive differential evolution algorithm

Transportation is a key issue in supply chain management and is a major concern for a company. This paper considers a joint-location inventory problem involving a set of suppliers producing different products and a set of retailers where some retailers are treated as distribution centers (DCs). The problem is to determine which retailers to be assigned as DCs, which retailers to receive direct shipments, how much of the retailer’s demand to allocate to the DCs, and how much of the DC’s demand is to be met by different suppliers. The problem is formulated as a mixed integer model and it has been solved through an adaptive differential evolution algorithm known as modified J. Adaptive Differential Evolution. The solutions obtained are compared with that of simple genetic algorithm. This paper also shows that the proposed model is robust in nature and offers near-optimal results for different distributions. The sum of the cost of establishing some retailers as DCs and the total transportation cost incurred in shipping products from the suppliers to the retailers via DCs(for some retailers) or directly (for the other retailers) is also compared with the total transportation cost incurred when all the products are shipped directly from the suppliers to the retailers.     

A replenishment policy for perishable products with non-linear holding cost under stochastic supply lead time

The development and application of inventory models for deteriorating items is one of the main concerns of subject matter experts. The inventory models developed in this field have focused mainly on supply chains under the assumption of constant lead time. In this study, we develop an inventory model for a main class of deteriorating items, namely perishable products, under stochastic lead time assumption. The inventory system is modeled as a continuous review system (r, Q). Demand rate per unit time is assumed to be constant over an infinite planning horizon and the shortages could be backordered completely. For modeling the deterioration process, a non-linear holding cost is considered. Taking into account the stochastic lead time as well as a non-linear holding cost makes the mathematical model more complicated. We customize the proposed model for a uniform distribution function that could be tractable to solve optimally by means of an exact approach. We then solve an example taken from the literature to demonstrate the applicability and effectiveness of the proposed model. Finally, by doing several sensitivity analyses for the key parameters of the model, some managerial insights are proposed.     

基于多层分解法和遗传算法的动态智能补货模型

为提高协同补货的高效性和智能性,运用多层分解法模型,建立了基于多智能体的分销链协同补货模型.在借鉴前人库存成本优化理论的基础上,以需求预测更新、惩罚或激励机制为前提,给出了基于传统规划理论的动态补货Agent和基于不同权重的协同补货Agent.案例分析了以补货量为变量的分销链两层公司广日j的补货期望成本优化问题,并运用遗传算法进行模拟,验证了动态补货Agent中的随机需求变化和协同补货Agent中的权重变化对分销链补货量和补货成本的动态影响.该模型和仿真结果表明,基于多智能体的动态协同补货模型具有动态性、交瓦性和智能性的特性,是有效可行的.     

单制造商-单销售商供应链的多物品联合补货及协调策略

为降低原材料制造商的成本,在同时考虑制造商的原材料固定订购费用、固定生产装配费用和销售商固定订购费用的情况下,研究了由单制造商和单销售商组成的两层供应链的联合补货问题.为揭示企业合作和多物品联合生产补货对供应链各成员及链的总费用的影响,建立了三种生产-补货模型,分析了模型的最优解特性,并给出了搜索算法.最后,给出一个延迟支付协调策略,实现了Pareto改进.     

一种新的求解多目标随机需求车辆路径问题的算法

为优化带时间窗的随机需求车辆路径问题,建立了基于模糊满意度的多目标数学规划模型,并提出了一种基于量子进化算法和粒子群算法分段优化的方法求解Pareto解。第一阶段使用量子进化算法获得一定规模和精度的Pareto候选解,提出了概率选择最优解和可变旋转角改进变异算子;第二阶段通过转换将候选解映射到连续空间,利用粒子群算法继续搜索Pareto最优解。引入了节点交换策略进行邻域搜索,避免算法早熟。为保持Pareto解的分散性,提出了一种自适应网格算子。通过对benchmark仿真与非支配排序的遗传算法的比较,验证显示了算法的有效性。